[ { "memory": 3058188544, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3060547840, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3061235968, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3069165824, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3071525120, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3079454976, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3087384832, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3095314688, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3103244544, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3105603840, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3113533696, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3115892992, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3123822848, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3131752704, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3139682560, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3147612416, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3149971712, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3150659840, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3158589696, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3166519552, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3167207680, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3175137536, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3183067392, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3190997248, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3191685376, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3192373504, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3193061632, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3195420928, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3203350784, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3211280640, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3211968768, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3212656896, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3213345024, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3221274880, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3229204736, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3237134592, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3239493888, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3247423744, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3255353600, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3256041728, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3258401024, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3266330880, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3268690176, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3269378304, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3270066432, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3270754560, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3273113856, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3275473152, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3277832448, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3280191744, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3288121600, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3296051456, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3296739584, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3299098880, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3307028736, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3307716864, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3310076160, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3310764288, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3318694144, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3326624000, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3327312128, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3335241984, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3337601280, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3339960576, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3347890432, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3350249728, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3358179584, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3366109440, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3374039296, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3381969152, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3389899008, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3397828864, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3398516992, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3399205120, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3401564416, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3403923712, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3406283008, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3414212864, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3416572160, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3424502016, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3426861312, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3434791168, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3442721024, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3443409152, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3444097280, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3452027136, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3452715264, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3460645120, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3463004416, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3470934272, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3473293568, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3473981696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3481911552, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3489841408, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3490529536, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3492888832, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3500818688, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3508748544, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3511107840, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3519037696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3526967552, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3534897408, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3542827264, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3550757120, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3553116416, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3561046272, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3561734400, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3564093696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3564781824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3565469952, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3573399808, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3581329664, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3583688960, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3584377088, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3585065216, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3592995072, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3593683200, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3596042496, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3598401792, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3599089920, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3601449216, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3602137344, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3602825472, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3610755328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3613114624, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3621044480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3623403776, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3625763072, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3633692928, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3634381056, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3636740352, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3644670208, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3645358336, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3653288192, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3653976320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3661906176, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3662594304, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3663282432, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3665641728, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3668001024, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3675930880, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3676619008, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3677307136, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3677995264, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3685925120, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3693854976, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3696214272, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3704144128, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3706503424, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3714433280, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3715121408, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3717480704, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3718168832, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3726098688, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3726786816, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3734716672, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3737075968, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3737764096, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3740123392, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3740811520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3743170816, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3745530112, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3746218240, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3754148096, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3754836224, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3757195520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3757883648, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3760242944, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3768172800, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3776102656, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3784032512, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3786391808, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3788751104, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3789439232, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3791798528, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3794157824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3794845952, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3797205248, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3805135104, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3807494400, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3809853696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3810541824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3818471680, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3826401536, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3834331392, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3835019520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3835707648, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3838066944, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3838755072, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3839443200, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3847373056, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 3855302912, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 2 } }, { "memory": 3863232768, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 2 } }, { "memory": 3865592064, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 2 } }, { "memory": 3873521920, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 2 } }, { "memory": 3875881216, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 2 } }, { "memory": 3883811072, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 2 } }, { "memory": 3891740928, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3899670784, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3902030080, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3909959936, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3917889792, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3918577920, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3926507776, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3927195904, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3927884032, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3930243328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3932602624, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3940532480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3941220608, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3943579904, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3951509760, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3959439616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3967369472, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3975299328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3975987456, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3983917312, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3991847168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3994206464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3994894592, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3995582720, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4003512576, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4004200704, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4012130560, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4014489856, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4015177984, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4017537280, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4025467136, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4033396992, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4041326848, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4043686144, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4051616000, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4053975296, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4061905152, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4064264448, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4072194304, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4074553600, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4082483456, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4083171584, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4091101440, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4099031296, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4106961152, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4114891008, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4122820864, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4130750720, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4138680576, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4146610432, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4154540288, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4156899584, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4159258880, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4159947008, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4160635136, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4168564992, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4170924288, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4178854144, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4181213440, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4183572736, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4185932032, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4188291328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4190650624, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4191338752, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4199268608, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4207198464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4215128320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4217487616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4225417472, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4233347328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4234035456, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4236394752, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4244324608, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4252254464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4254613760, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4256973056, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4257661184, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4265591040, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4267950336, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4275880192, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4276568320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4278927616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4279615744, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4287545600, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4295475456, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4303405312, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4304093440, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4304781568, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4312711424, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4313399552, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4315758848, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4323688704, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4331618560, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4339548416, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4347478272, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4348166400, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4356096256, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4364026112, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4371955968, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4374315264, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4382245120, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4382933248, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4383621376, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 4391551232, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4393910528, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4401840384, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4409770240, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4410458368, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4418388224, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4420747520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4423106816, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4431036672, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4431724800, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4439654656, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4447584512, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4449943808, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4457873664, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4460232960, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4462592256, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4463280384, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4471210240, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4479140096, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4479828224, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4487758080, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4488446208, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4490805504, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4493164800, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4493852928, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4501782784, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4509712640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4517642496, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4520001792, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4520689920, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4523049216, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4523737344, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4526096640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4526784768, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4529144064, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4537073920, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4537762048, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4540121344, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4540809472, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4541497600, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4542185728, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4550115584, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4550803712, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4558733568, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4566663424, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4567351552, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4569710848, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4577640704, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4585570560, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4586258688, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4594188544, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4602118400, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4610048256, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4612407552, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4614766848, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4615454976, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4616143104, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4616831232, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4624761088, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4625449216, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4627808512, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4635738368, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4643668224, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4646027520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4648386816, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4656316672, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4658675968, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4666605824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4668965120, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4669653248, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4672012544, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4672700672, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4673388800, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4681318656, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4682006784, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4689936640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4690624768, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4691312896, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4693672192, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4696031488, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4696719616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4704649472, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4712579328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4720509184, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4728439040, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4729127168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4737057024, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4739416320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4741775616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4744134912, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4752064768, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4752752896, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4753441024, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4761370880, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4769300736, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4769988864, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4770676992, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4778606848, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4786536704, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4794466560, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4796825856, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4799185152, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4801544448, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4802232576, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4802920704, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4805280000, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4807639296, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4815569152, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4823499008, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4825858304, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4833788160, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4834476288, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4835164416, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4843094272, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4843782400, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4844470528, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4846829824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4847517952, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4848206080, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4848894208, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4851253504, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4859183360, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4861542656, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4869472512, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4870160640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4870848768, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4873208064, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4873896192, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4876255488, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4884185344, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4886544640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4894474496, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4896833792, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4904763648, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4912693504, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4915052800, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4917412096, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4918100224, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4918788352, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4926718208, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4929077504, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4937007360, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4944937216, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4952867072, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4960796928, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4961485056, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 4969414912, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 4970103040, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 4970791168, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 4971479296, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 4972167424, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 4980097280, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 4988027136, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 4995956992, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5003886848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5011816704, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5019746560, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5022105856, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5030035712, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5037965568, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5038653696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5041012992, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5048942848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5049630976, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5057560832, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5065490688, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5073420544, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5075779840, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5083709696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 5091639552, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5093998848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5096358144, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5098717440, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5106647296, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5114577152, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5122507008, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5130436864, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5138366720, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5146296576, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5148655872, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5156585728, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5164515584, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5165203712, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5173133568, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5173821696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5181751552, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5184110848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5186470144, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5194400000, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5202329856, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5210259712, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5218189568, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5220548864, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5228478720, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5229166848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5231526144, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5239456000, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5241815296, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5242503424, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5243191552, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5251121408, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5259051264, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5266981120, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5274910976, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5277270272, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5277958400, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5280317696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5288247552, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5296177408, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5304107264, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5312037120, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5319966976, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5327896832, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5328584960, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5330944256, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5338874112, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5346803968, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5354733824, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5362663680, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5370593536, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5378523392, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5380882688, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5388812544, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5396742400, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5404672256, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5412602112, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5413290240, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5421220096, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5423579392, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5431509248, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5439439104, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5447368960, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5455298816, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5463228672, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5463916800, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5471846656, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5479776512, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5480464640, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5482823936, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5483512064, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5491441920, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5499371776, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5501731072, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5502419200, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5504778496, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5505466624, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5507825920, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5510185216, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5518115072, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5526044928, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5528404224, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5536334080, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5544263936, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5552193792, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5560123648, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5568053504, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5568741632, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5576671488, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5579030784, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5586960640, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5587648768, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5588336896, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5590696192, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5598626048, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5600985344, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5603344640, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5611274496, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5619204352, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5621563648, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5629493504, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5637423360, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5638111488, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5640470784, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5642830080, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5643518208, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5645877504, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5653807360, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5654495488, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5655183616, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5663113472, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5665472768, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5667832064, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5675761920, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5678121216, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5678809344, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5686739200, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5687427328, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5695357184, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5696045312, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5696733440, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5697421568, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5698109696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5700468992, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5708398848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5709086976, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5717016832, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5717704960, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5720064256, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5720752384, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5721440512, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5722128640, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5724487936, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5726847232, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5727535360, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5728223488, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5730582784, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5732942080, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5735301376, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5737660672, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5738348800, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5746278656, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5754208512, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5754896640, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5755584768, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5757944064, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5758632192, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5766562048, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5774491904, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5775180032, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5777539328, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5785469184, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5786157312, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5786845440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5789204736, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5789892864, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5792252160, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5800182016, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5800870144, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5808800000, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5811159296, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5813518592, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5815877888, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 4 } }, { "memory": 5823807744, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5824495872, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5826855168, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5829214464, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5829902592, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5830590720, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5832950016, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5833638144, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5835997440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5836685568, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5844615424, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5852545280, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5853233408, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5855592704, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5856280832, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5858640128, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5859328256, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5861687552, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5862375680, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5863063808, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5865423104, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5867782400, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5868470528, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5876400384, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5878759680, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5881118976, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5883478272, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5891408128, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5893767424, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5894455552, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5895143680, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5895831808, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5903761664, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5904449792, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5905137920, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5905826048, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5906514176, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5908873472, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5911232768, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5919162624, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5927092480, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 5935022336, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5937381632, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5938069760, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5945999616, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5953929472, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5956288768, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5956976896, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5964906752, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5972836608, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5973524736, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5974212864, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5982142720, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5990072576, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 5998002432, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6005932288, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6013862144, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6016221440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6018580736, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6026510592, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6028869888, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6036799744, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6044729600, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6052659456, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6055018752, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6062948608, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6070878464, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6078808320, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6086738176, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6094668032, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6102597888, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6110527744, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6118457600, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6120816896, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6128746752, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6136676608, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6144606464, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6152536320, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6153224448, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6153912576, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6161842432, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6169772288, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6177702144, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6185632000, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6187991296, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6195921152, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6203851008, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6211780864, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6219710720, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6222070016, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6224429312, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6232359168, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6240289024, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6240977152, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6248907008, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6256836864, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6264766720, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6272696576, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6280626432, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6288556288, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6296486144, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6304416000, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6312345856, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6320275712, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6322635008, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6324994304, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6327353600, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6335283456, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6335971584, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6343901440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6346260736, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6354190592, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6362120448, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6370050304, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6377980160, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6385910016, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6388269312, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6396199168, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6404129024, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6412058880, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6412747008, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6413435136, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6421364992, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6423724288, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6426083584, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6434013440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6434701568, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6437060864, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6439420160, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6440108288, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6448038144, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6450397440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6451085568, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6459015424, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6466945280, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6474875136, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6482804992, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6490734848, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6491422976, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6499352832, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6507282688, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6507970816, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6508658944, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6516588800, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6524518656, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6525206784, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6527566080, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6529925376, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6537855232, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6538543360, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6539231488, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6541590784, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6542278912, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6550208768, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6552568064, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6560497920, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6568427776, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6570787072, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6571475200, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6572163328, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6572851456, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6580781312, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6581469440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6589399296, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6597329152, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6598017280, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6598705408, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6606635264, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6608994560, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6609682688, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6617612544, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6619971840, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6622331136, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6624690432, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6632620288, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6633308416, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6635667712, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6636355840, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6637043968, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6637732096, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6645661952, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6648021248, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6655951104, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6656639232, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6657327360, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6659686656, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6667616512, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6675546368, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6677905664, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6680264960, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6688194816, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6690554112, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6698483968, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6706413824, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6708773120, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6709461248, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6710149376, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6712508672, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6720438528, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6721126656, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6723485952, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6724174080, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6724862208, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6727221504, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6729580800, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6730268928, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6730957056, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6731645184, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6732333312, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6733021440, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6740951296, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6743310592, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6743998720, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6751928576, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6754287872, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6756647168, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6759006464, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6761365760, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6763725056, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6771654912, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6779584768, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6781944064, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6782632192, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6790562048, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6798491904, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6799180032, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6801539328, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6802227456, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6804586752, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6812516608, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6813204736, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6815564032, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6816252160, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6824182016, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6824870144, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6825558272, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6827917568, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6828605696, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6830964992, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6831653120, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6834012416, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6836371712, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6838731008, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6839419136, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6840107264, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6840795392, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6843154688, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6843842816, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6846202112, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6848561408, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6850920704, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6851608832, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6852296960, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6854656256, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6857015552, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6857703680, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6858391808, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6859079936, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6861439232, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6863798528, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6871728384, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6872416512, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 5 } }, { "memory": 6880346368, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 6882705664, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 6883393792, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 6891323648, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 6 } }, { "memory": 6899253504, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6899941632, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6900629760, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6908559616, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6916489472, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6924419328, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6932349184, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6933037312, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6940967168, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6948897024, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6956826880, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6964756736, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6972686592, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6980616448, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6988546304, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6996476160, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6997164288, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7005094144, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7013024000, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7020953856, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7021641984, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7022330112, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7023018240, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7030948096, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7031636224, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7039566080, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7047495936, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7055425792, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7063355648, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7071285504, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7073644800, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7076004096, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7078363392, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7079051520, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7086981376, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7087669504, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7088357632, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7096287488, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7104217344, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7106576640, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7114506496, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7122436352, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7130366208, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7138296064, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7140655360, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7148585216, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7156515072, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7164444928, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7172374784, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7173062912, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7180992768, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7188922624, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7196852480, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7197540608, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7205470464, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7213400320, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7214088448, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7222018304, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7229948160, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7237878016, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7245807872, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7253737728, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7261667584, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7269597440, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7277527296, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7279886592, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7287816448, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7295746304, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7303676160, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7311606016, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7319535872, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7327465728, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7335395584, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7337754880, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7345684736, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7353614592, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7355973888, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7363903744, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7371833600, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7379763456, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7380451584, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7381139712, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7389069568, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7391428864, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7393788160, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7396147456, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7404077312, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7412007168, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7419937024, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7422296320, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7430226176, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7438156032, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7446085888, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7448445184, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7449133312, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7449821440, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7457751296, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7465681152, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7468040448, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7475970304, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7478329600, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7479017728, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7479705856, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7487635712, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7489995008, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7490683136, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7498612992, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7506542848, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7514472704, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7516832000, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7519191296, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7521550592, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7523909888, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7526269184, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7526957312, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7529316608, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7537246464, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7539605760, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7547535616, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7555465472, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7556153600, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7558512896, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7559201024, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7559889152, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7560577280, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7568507136, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7576436992, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7584366848, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7592296704, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7594656000, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7597015296, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7604945152, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7607304448, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7607992576, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7608680704, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7611040000, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7618969856, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7619657984, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7620346112, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7622705408, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7623393536, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7624081664, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7624769792, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7632699648, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7635058944, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7635747072, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7643676928, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7644365056, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7646724352, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7647412480, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7649771776, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7650459904, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7652819200, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7653507328, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7654195456, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7656554752, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7664484608, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7665172736, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7673102592, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7673790720, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7676150016, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7676838144, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7684768000, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7692697856, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7695057152, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7702987008, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7703675136, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7706034432, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7706722560, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7707410688, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7709769984, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7710458112, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7712817408, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7715176704, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7717536000, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7719895296, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7720583424, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7721271552, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7723630848, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7725990144, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7728349440, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7736279296, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7738638592, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7740997888, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7743357184, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7744045312, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7744733440, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7752663296, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7753351424, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7761281280, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7763640576, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7764328704, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7772258560, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7774617856, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7776977152, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7777665280, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7780024576, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7780712704, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7781400832, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7782088960, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7782777088, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7785136384, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7785824512, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7788183808, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7788871936, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7789560064, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7791919360, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7794278656, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7794966784, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7795654912, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7803584768, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7805944064, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7806632192, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7808991488, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7811350784, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7819280640, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7827210496, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7829569792, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7831929088, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7834288384, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7836647680, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7839006976, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7839695104, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7840383232, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7841071360, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7849001216, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 7856931072, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7864860928, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7867220224, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7875150080, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7877509376, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7885439232, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7893369088, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7894057216, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7896416512, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7897104640, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7905034496, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7912964352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7915323648, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7916011776, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7918371072, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7926300928, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7934230784, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7934918912, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7937278208, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7937966336, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7945896192, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7953826048, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7956185344, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7964115200, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7972045056, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7974404352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7982334208, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7984693504, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 7992623360, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8000553216, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8008483072, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8016412928, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8024342784, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8032272640, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8034631936, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8042561792, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8050491648, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8058421504, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8066351360, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8074281216, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8074969344, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8082899200, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8085258496, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8093188352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8101118208, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8101806336, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8109736192, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8117666048, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 8125595904, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8133525760, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8135885056, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8143814912, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8151744768, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8159674624, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8160362752, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8168292608, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8170651904, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8178581760, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8186511616, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8194441472, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8202371328, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8203059456, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8210989312, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8218919168, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8226849024, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8229208320, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8237138176, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8239497472, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8247427328, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8248115456, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8256045312, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8263975168, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8266334464, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8274264320, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8274952448, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8282882304, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8290812160, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8298742016, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8306671872, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8314601728, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8322531584, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8330461440, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8338391296, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8340750592, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8343109888, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8343798016, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8344486144, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8352416000, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8360345856, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8361033984, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8368963840, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8376893696, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8379252992, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8379941120, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8380629248, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8388559104, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8389247232, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8389935360, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8392294656, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8400224512, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8408154368, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8416084224, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8416772352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8417460480, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8425390336, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8433320192, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8434008320, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8434696448, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8437055744, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8444985600, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8445673728, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8446361856, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8454291712, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8456651008, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8464580864, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8472510720, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8480440576, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8482799872, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8485159168, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8485847296, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8488206592, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8488894720, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8496824576, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8497512704, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8505442560, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8506130688, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8514060544, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8521990400, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8529920256, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8530608384, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8532967680, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8535326976, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8537686272, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8545616128, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8547975424, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8555905280, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8558264576, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8558952704, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8561312000, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8563671296, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8564359424, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8565047552, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8572977408, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8573665536, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8581595392, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8582283520, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8582971648, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8590901504, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8598831360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8601190656, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8601878784, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8602566912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8604926208, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8612856064, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8615215360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8623145216, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8625504512, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8633434368, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8641364224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8643723520, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8644411648, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8645099776, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8647459072, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8648147200, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8650506496, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8652865792, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8653553920, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8661483776, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8669413632, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8671772928, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8672461056, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8674820352, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8675508480, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8676196608, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8678555904, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8680915200, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8688845056, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8691204352, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8693563648, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8695922944, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8696611072, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8698970368, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8701329664, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8709259520, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8711618816, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8719548672, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8720236800, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8722596096, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8723284224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8725643520, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8726331648, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8734261504, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8736620800, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8738980096, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8741339392, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8743698688, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8744386816, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8745074944, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8747434240, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8748122368, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8756052224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8756740352, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8757428480, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8758116608, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8766046464, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8768405760, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8769093888, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8771453184, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8772141312, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8774500608, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8776859904, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8777548032, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8778236160, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8780595456, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8781283584, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8789213440, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8791572736, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8792260864, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8794620160, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8795308288, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8795996416, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 8803926272, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8804614400, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8805302528, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8813232384, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8821162240, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8821850368, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8829780224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8837710080, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8845639936, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8853569792, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8861499648, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8869429504, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8877359360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8885289216, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8887648512, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8895578368, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8897937664, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8905867520, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8913797376, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8916156672, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8924086528, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8932016384, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8939946240, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8940634368, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8948564224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8956494080, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8958853376, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8966783232, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8967471360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8975401216, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8983331072, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8991260928, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 8993620224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9001550080, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9009479936, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9017409792, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9025339648, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9033269504, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9041199360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9041887488, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9049817344, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9057747200, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9065677056, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9066365184, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9074295040, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9082224896, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9090154752, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9098084608, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9100443904, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9108373760, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9109061888, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9111421184, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9112109312, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9120039168, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9127969024, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9135898880, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9143828736, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9151758592, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9159688448, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9167618304, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9168306432, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9170665728, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9178595584, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9186525440, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9187213568, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9189572864, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9190260992, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9190949120, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9198878976, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9206808832, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9214738688, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9222668544, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9230598400, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9231286528, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9231974656, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9239904512, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9240592640, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9248522496, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9250881792, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9258811648, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9261170944, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9263530240, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9271460096, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9272148224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9280078080, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9288007936, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9290367232, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9291055360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9298985216, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9306915072, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9309274368, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9317204224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9325134080, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9333063936, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9333752064, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9336111360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9344041216, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9351971072, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9359900928, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9360589056, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9368518912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9370878208, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9378808064, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9381167360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9381855488, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9382543616, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9384902912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9392832768, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9393520896, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9394209024, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9394897152, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9395585280, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9403515136, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9405874432, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9413804288, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9416163584, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9424093440, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9426452736, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9427140864, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9427828992, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9435758848, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9443688704, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9446048000, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9453977856, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9461907712, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9462595840, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9464955136, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9472884992, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9475244288, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9483174144, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9485533440, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9486221568, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9494151424, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9496510720, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9498870016, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9499558144, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9501917440, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9504276736, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9506636032, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9507324160, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9509683456, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9512042752, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9512730880, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9513419008, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9521348864, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9522036992, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9522725120, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9523413248, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9525772544, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9526460672, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9528819968, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9529508096, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9530196224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9538126080, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9546055936, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9546744064, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9554673920, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9557033216, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9559392512, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9561751808, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9564111104, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9564799232, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9565487360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9566175488, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9568534784, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9570894080, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9573253376, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9575612672, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9577971968, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9580331264, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9588261120, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9590620416, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 8 } }, { "memory": 9598550272, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9606480128, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9608839424, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9611198720, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9613558016, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9621487872, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9623847168, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9631777024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9639706880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9640395008, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9642754304, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9645113600, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9645801728, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9648161024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9648849152, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9656779008, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9657467136, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9658155264, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9660514560, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9661202688, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9663561984, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9664250112, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9666609408, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9674539264, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9682469120, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9684828416, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9685516544, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9686204672, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9688563968, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9689252096, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9689940224, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9690628352, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9692987648, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9700917504, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9701605632, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9709535488, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9711894784, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9712582912, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9713271040, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9721200896, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9729130752, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9731490048, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9733849344, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9736208640, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9736896768, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9737584896, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9738273024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9746202880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9748562176, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9749250304, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9749938432, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9752297728, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9752985856, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9753673984, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9754362112, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9756721408, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9764651264, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9765339392, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 9773269248, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9775628544, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9776316672, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9778675968, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9786605824, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9794535680, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9802465536, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9810395392, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9818325248, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9826255104, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9834184960, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9842114816, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9850044672, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9857974528, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9858662656, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9866592512, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9874522368, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9882452224, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9890382080, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9898311936, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9900671232, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9908601088, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9916530944, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9924460800, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9932390656, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9940320512, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9948250368, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9956180224, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9964110080, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9964798208, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9967157504, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9975087360, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9977446656, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9978134784, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9978822912, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9986752768, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9989112064, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9989800192, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9997730048, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 9998418176, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10006348032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10008707328, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10009395456, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10017325312, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10025255168, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10033185024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10041114880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10049044736, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10056974592, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10064904448, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10067263744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10075193600, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10083123456, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10085482752, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10086170880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10094100736, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10102030592, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10109960448, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10117890304, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10118578432, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10126508288, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10134438144, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10135126272, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10143056128, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10143744256, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10151674112, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10154033408, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10161963264, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10169893120, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10177822976, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10180182272, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10182541568, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10183229696, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10191159552, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10199089408, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10201448704, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10203808000, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10206167296, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10206855424, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10207543552, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10215473408, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10223403264, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10231333120, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10232021248, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10239951104, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10242310400, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10250240256, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10250928384, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10251616512, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10252304640, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10254663936, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10262593792, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10263281920, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10265641216, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10273571072, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10281500928, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10289430784, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10297360640, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10298048768, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10305978624, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10313908480, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10321838336, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10322526464, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10323214592, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10325573888, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10333503744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10335863040, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10338222336, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10338910464, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10339598592, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10340286720, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10348216576, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10356146432, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10364076288, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10366435584, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10368794880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10371154176, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 10379084032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10387013888, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10394943744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10395631872, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10397991168, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10405921024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10408280320, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10410639616, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10412998912, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10420928768, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10421616896, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10429546752, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10430234880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10438164736, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10440524032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10442883328, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10445242624, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10447601920, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10449961216, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10457891072, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10465820928, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10468180224, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10468868352, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10471227648, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10471915776, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10474275072, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10474963200, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10475651328, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10476339456, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10478698752, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10486628608, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10488987904, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10489676032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10490364160, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10492723456, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10493411584, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10494099712, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10502029568, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10502717696, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10503405824, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10511335680, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10512023808, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10514383104, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10515071232, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10517430528, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10519789824, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10520477952, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10522837248, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10525196544, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10533126400, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10535485696, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10537844992, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10538533120, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10540892416, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10543251712, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10545611008, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10546299136, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10546987264, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10547675392, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10555605248, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10563535104, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10564223232, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10564911360, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10565599488, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10567958784, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10575888640, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10583818496, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10584506624, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10592436480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10594795776, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10597155072, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10599514368, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10601873664, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10602561792, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10604921088, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10605609216, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10606297344, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10614227200, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10616586496, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10618945792, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10621305088, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10629234944, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10637164800, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10637852928, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10638541056, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10640900352, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10648830208, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10656760064, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10657448192, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10659807488, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10667737344, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10670096640, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10672455936, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10673144064, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10681073920, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10689003776, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10696933632, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10699292928, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10707222784, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10707910912, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10708599040, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 10716528896, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 10 } }, { "memory": 10724458752, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 10 } }, { "memory": 10732388608, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 10 } }, { "memory": 10733076736, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 10 } }, { "memory": 10733764864, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 10 } }, { "memory": 10741694720, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 10 } }, { "memory": 10749624576, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10757554432, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10765484288, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10766172416, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10774102272, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10782032128, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10789961984, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10797891840, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10805821696, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10806509824, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10808869120, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10811228416, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10819158272, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10827088128, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10835017984, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10842947840, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10850877696, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10858807552, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10859495680, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10867425536, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10868113664, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10876043520, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10878402816, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10879090944, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10887020800, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10894950656, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10902880512, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10903568640, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10911498496, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10919428352, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10921787648, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10929717504, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10937647360, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10945577216, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10953507072, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10955866368, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10963796224, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10971726080, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10974085376, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10974773504, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10975461632, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10983391488, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10991321344, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10999251200, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11007181056, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11007869184, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11010228480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11018158336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11026088192, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11028447488, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11036377344, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11037065472, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11039424768, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11047354624, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11055284480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11063214336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11071144192, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11071832320, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11074191616, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11082121472, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11082809600, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11085168896, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11093098752, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11101028608, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11101716736, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11109646592, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11112005888, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11114365184, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11116724480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11117412608, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11125342464, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11126030592, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11126718720, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11127406848, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11128094976, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11128783104, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11136712960, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11137401088, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11138089216, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11138777344, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11146707200, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11154637056, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11162566912, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11170496768, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11171184896, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11171873024, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11179802880, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11180491008, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11188420864, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11196350720, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11198710016, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11201069312, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11203428608, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11204116736, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11204804864, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11212734720, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11215094016, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11217453312, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11219812608, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11222171904, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11222860032, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11230789888, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11231478016, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11233837312, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11234525440, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11242455296, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11243143424, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11245502720, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11247862016, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11248550144, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11250909440, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11258839296, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11266769152, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11269128448, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11269816576, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11277746432, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11280105728, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11280793856, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11283153152, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11291083008, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11291771136, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11294130432, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11302060288, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11309990144, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11310678272, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11318608128, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11320967424, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11321655552, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11329585408, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11337515264, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11345445120, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11353374976, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11355734272, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11356422400, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11358781696, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11361140992, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11369070848, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11371430144, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11372118272, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11380048128, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11380736256, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11388666112, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11389354240, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11391713536, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11399643392, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11407573248, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11415503104, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11417862400, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11425792256, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11433722112, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11434410240, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11436769536, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11439128832, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11439816960, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11447746816, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11448434944, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11450794240, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11458724096, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11466653952, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11467342080, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11468030208, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11468718336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11471077632, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11479007488, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11481366784, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11483726080, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11484414208, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11492344064, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11493032192, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11500962048, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11501650176, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11504009472, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11504697600, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11507056896, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11507745024, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11510104320, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11518034176, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11520393472, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11522752768, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11525112064, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11525800192, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11528159488, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11536089344, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11538448640, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11540807936, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11543167232, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11543855360, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11546214656, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11554144512, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11556503808, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11557191936, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11565121792, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11573051648, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11575410944, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11577770240, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11585700096, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11588059392, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11588747520, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11596677376, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11597365504, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11598053632, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11598741760, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11601101056, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11603460352, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11605819648, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11608178944, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11610538240, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11618468096, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11626397952, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11634327808, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11635015936, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11635704064, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11643633920, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11651563776, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11653923072, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11654611200, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11662541056, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11664900352, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 11672830208, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11680760064, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11683119360, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11685478656, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11686166784, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11694096640, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11694784768, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11702714624, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11710644480, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11718574336, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11720933632, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11723292928, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11731222784, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11731910912, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11732599040, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11734958336, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11742888192, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11750818048, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11758747904, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 10 } }, { "memory": 11766677760, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11774607616, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11775295744, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11783225600, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11783913728, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11791843584, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11792531712, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11794891008, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11795579136, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11803508992, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11811438848, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11819368704, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11827298560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11835228416, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11843158272, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11851088128, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11859017984, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11866947840, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11874877696, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11882807552, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11890737408, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11898667264, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11906597120, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11914526976, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11922456832, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11924816128, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11932745984, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11940675840, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11948605696, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11950964992, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11958894848, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11961254144, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11969184000, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11977113856, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11977801984, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11980161280, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11988091136, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11988779264, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11996709120, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11999068416, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12006998272, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12007686400, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12015616256, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12023546112, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12024234240, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12032164096, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12040093952, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12048023808, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12055953664, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12056641792, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12059001088, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12066930944, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12074860800, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12077220096, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12085149952, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12087509248, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12095439104, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12096127232, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12104057088, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12104745216, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12105433344, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12107792640, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12110151936, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12118081792, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12126011648, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12128370944, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12130730240, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12138660096, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12146589952, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12154519808, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12155207936, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12163137792, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12171067648, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12173426944, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12181356800, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12189286656, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12197216512, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12199575808, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12200263936, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12200952064, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12208881920, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12216811776, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12217499904, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12225429760, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12227789056, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12235718912, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12238078208, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12240437504, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12241125632, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12241813760, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12244173056, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12246532352, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12247220480, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12255150336, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12257509632, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12258197760, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12266127616, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12266815744, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12274745600, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12282675456, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12283363584, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12285722880, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12288082176, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12290441472, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12298371328, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12299059456, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12301418752, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12303778048, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12306137344, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12306825472, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12309184768, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12311544064, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12312232192, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12312920320, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12320850176, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12323209472, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12331139328, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12339069184, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12341428480, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12343787776, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12344475904, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12345164032, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12345852160, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12346540288, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12354470144, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12356829440, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12357517568, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12365447424, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12367806720, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12375736576, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12383666432, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12384354560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12385042688, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12387401984, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12388090112, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12390449408, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12398379264, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12400738560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12401426688, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12403785984, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12411715840, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12412403968, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12414763264, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12417122560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12417810688, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12418498816, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12426428672, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12434358528, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12436717824, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12437405952, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12438094080, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12446023936, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12446712064, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12454641920, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12455330048, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12456018176, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12456706304, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12457394432, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12459753728, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12460441856, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12468371712, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12470731008, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12478660864, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12481020160, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12481708288, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12482396416, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12484755712, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12492685568, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12495044864, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12495732992, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12498092288, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12498780416, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12499468544, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12507398400, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12515328256, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12523258112, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12523946240, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12526305536, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12528664832, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12529352960, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12531712256, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12532400384, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12533088512, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12535447808, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12537807104, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12540166400, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12540854528, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12541542656, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12543901952, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12551831808, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12554191104, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12562120960, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12564480256, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12566839552, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12574769408, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12575457536, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12583387392, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12585746688, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12586434816, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12594364672, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12602294528, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12610224384, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12618154240, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12620513536, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12621201664, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 12629131520, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 12 } }, { "memory": 12637061376, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 12 } }, { "memory": 12644991232, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12652921088, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12655280384, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12655968512, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12663898368, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12671828224, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12679758080, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12687687936, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12688376064, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12689064192, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12696994048, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12699353344, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12701712640, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12709642496, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12712001792, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12712689920, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12720619776, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12728549632, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12736479488, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12744409344, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12752339200, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12754698496, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12762628352, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12763316480, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12765675776, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12766363904, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12774293760, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12782223616, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12790153472, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12792512768, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12794872064, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12797231360, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12805161216, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12813091072, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12821020928, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12828950784, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12836880640, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12844810496, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12852740352, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12860670208, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12868600064, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12869288192, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12871647488, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12872335616, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12880265472, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12882624768, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12890554624, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12898484480, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12906414336, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12907102464, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12915032320, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12922962176, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12925321472, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12926009600, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12933939456, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12941869312, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12944228608, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12952158464, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12952846592, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12953534720, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12961464576, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12969394432, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12977324288, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12985254144, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12987613440, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12988301568, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12996231424, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12996919552, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12999278848, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13007208704, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13007896832, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13008584960, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13016514816, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13024444672, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13026803968, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13027492096, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13029851392, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13037781248, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13040140544, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13042499840, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13043187968, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13051117824, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13051805952, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13054165248, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13056524544, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13057212672, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13057900800, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13065830656, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13066518784, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13074448640, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13076807936, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13077496064, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13085425920, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13086114048, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13088473344, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13089161472, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13097091328, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13099450624, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13101809920, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13109739776, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13110427904, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13111116032, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13113475328, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13114163456, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13122093312, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13124452608, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13126811904, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13127500032, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13128188160, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13128876288, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13131235584, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13131923712, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13139853568, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13140541696, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13142900992, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13150830848, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13153190144, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13155549440, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13157908736, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13158596864, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13166526720, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13168886016, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13169574144, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13177504000, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13178192128, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13186121984, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13188481280, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13196411136, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13197099264, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13199458560, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13207388416, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13208076544, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13210435840, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13212795136, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13220724992, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13228654848, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13231014144, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13233373440, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13235732736, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13236420864, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13238780160, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13239468288, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13247398144, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13249757440, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13257687296, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13265617152, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13273547008, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13281476864, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13282164992, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13282853120, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13290782976, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13293142272, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13295501568, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13303431424, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13305790720, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13306478848, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13314408704, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13316768000, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13317456128, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13318144256, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13320503552, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13322862848, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13330792704, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13331480832, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13332168960, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13332857088, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13335216384, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13337575680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13338263808, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13346193664, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13346881792, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13354811648, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13362741504, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13370671360, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13373030656, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13380960512, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13381648640, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13389578496, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13397508352, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13399867648, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13407797504, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13408485632, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13416415488, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13418774784, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13421134080, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13421822208, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13424181504, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13424869632, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13427228928, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13427917056, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13435846912, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13443776768, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13446136064, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13454065920, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13461995776, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13469925632, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13470613760, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13478543616, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13486473472, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13487161600, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13487849728, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13490209024, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13492568320, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13500498176, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13508428032, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13510787328, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13513146624, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13521076480, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13521764608, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13529694464, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13537624320, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13539983616, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13540671744, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13541359872, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13549289728, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13557219584, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13557907712, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13565837568, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13566525696, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13567213824, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13575143680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13577502976, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13585432832, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13593362688, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13595721984, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13598081280, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13606011136, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13613940992, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13621870848, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13622558976, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13630488832, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13631176960, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13631865088, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13639794944, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13647724800, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13655654656, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13656342784, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13657030912, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13657719040, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13665648896, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13666337024, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13674266880, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13682196736, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13682884864, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13690814720, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13698744576, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13706674432, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13714604288, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13722534144, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13723222272, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13731152128, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13739081984, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13739770112, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13747699968, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13755629824, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13763559680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13771489536, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13779419392, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13787349248, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13789708544, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13797638400, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13799997696, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13807927552, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13808615680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13816545536, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13824475392, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13832405248, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13834764544, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13842694400, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13850624256, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13858554112, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 13866483968, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13874413824, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13882343680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13890273536, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13890961664, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13893320960, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13901250816, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13909180672, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13917110528, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13917798656, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13920157952, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13928087808, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13936017664, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13943947520, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13946306816, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13954236672, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13954924800, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13955612928, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13956301056, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13964230912, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13964919040, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13965607168, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13973537024, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13975896320, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13983826176, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13991756032, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 13999685888, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14002045184, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14009975040, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14010663168, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14018593024, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14020952320, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14028882176, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14029570304, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14030258432, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14032617728, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14033305856, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14041235712, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 13 } }, { "memory": 14049165568, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14049853696, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14057783552, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14058471680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14066401536, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14074331392, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14082261248, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14084620544, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14086979840, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14089339136, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14091698432, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14092386560, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14093074688, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14095433984, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14096122112, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14098481408, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14106411264, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14114341120, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14122270976, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14124630272, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14125318400, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14133248256, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14135607552, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14137966848, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14138654976, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14146584832, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14154514688, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14162444544, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14164803840, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14167163136, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14167851264, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14170210560, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14172569856, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14174929152, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14182859008, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14183547136, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14185906432, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14188265728, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14196195584, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14196883712, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14199243008, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14201602304, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14202290432, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14202978560, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14205337856, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14213267712, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14215627008, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14223556864, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14224244992, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14224933120, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14232862976, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14235222272, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14243152128, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14243840256, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14251770112, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14259699968, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14267629824, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14268317952, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14270677248, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14273036544, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14280966400, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14281654528, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14282342656, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14283030784, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14290960640, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14293319936, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14301249792, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14301937920, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14309867776, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14310555904, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14312915200, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14315274496, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14317633792, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14318321920, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14320681216, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14328611072, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14330970368, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14333329664, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14335688960, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14336377088, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14337065216, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14344995072, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14347354368, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14348042496, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14355972352, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14356660480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14359019776, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14361379072, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14369308928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14377238784, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14385168640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14387527936, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14389887232, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14397817088, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14400176384, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14408106240, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14410465536, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14418395392, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14419083520, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14421442816, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14422130944, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14424490240, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14425178368, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14425866496, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14426554624, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14434484480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14435172608, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14437531904, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14445461760, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14446149888, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14448509184, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14450868480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14451556608, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14459486464, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14460174592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14462533888, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14463222016, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14471151872, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14471840000, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14474199296, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14476558592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14484488448, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14492418304, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14494777600, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14497136896, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14497825024, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14500184320, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14502543616, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14510473472, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14518403328, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14520762624, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14523121920, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14523810048, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14524498176, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14525186304, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14533116160, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14533804288, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14534492416, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 14542422272, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14544781568, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14545469696, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14553399552, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14561329408, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14569259264, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14577189120, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14585118976, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14593048832, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 14600978688, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14608908544, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14616838400, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14624768256, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14627127552, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14627815680, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14628503808, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14630863104, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14631551232, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14639481088, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14640169216, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14648099072, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14656028928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14663958784, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14671888640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14672576768, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14680506624, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14682865920, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14690795776, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14693155072, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14701084928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14709014784, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14709702912, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14717632768, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14725562624, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14733492480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14735851776, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14743781632, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14746140928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14754070784, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14762000640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14769930496, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14777860352, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14780219648, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14780907776, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14788837632, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14796767488, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14804697344, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14805385472, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14807744768, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14815674624, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14823604480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14831534336, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14832222464, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14832910592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14833598720, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14835958016, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14843887872, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14846247168, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14848606464, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14856536320, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14857224448, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14865154304, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14873084160, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14875443456, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14876131584, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14876819712, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14884749568, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14892679424, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14893367552, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14894055680, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14896414976, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14898774272, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14901133568, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14901821696, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14909751552, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14910439680, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14912798976, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14920728832, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14928658688, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14931017984, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14938947840, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14939635968, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14941995264, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14949925120, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14952284416, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14954643712, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14955331840, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14956019968, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14958379264, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14966309120, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14968668416, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14971027712, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14973387008, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14975746304, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14976434432, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14977122560, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14977810688, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14985740544, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14988099840, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14988787968, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14991147264, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14991835392, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14992523520, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14994882816, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15002812672, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15005171968, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15013101824, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15013789952, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15016149248, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15016837376, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15024767232, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15025455360, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15027814656, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15035744512, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15036432640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15044362496, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15052292352, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15054651648, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15055339776, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15057699072, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15060058368, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15060746496, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15061434624, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15063793920, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15064482048, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15072411904, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15074771200, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15075459328, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15083389184, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15091319040, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15093678336, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15096037632, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15096725760, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15099085056, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15101444352, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15102132480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15102820608, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15103508736, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15111438592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15119368448, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15121727744, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15124087040, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15126446336, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15128805632, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15136735488, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15144665344, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15147024640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15154954496, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15155642624, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15163572480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15165931776, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15173861632, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15174549760, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15175237888, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15177597184, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15185527040, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15187886336, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15195816192, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15203746048, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15204434176, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15205122304, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15205810432, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15213740288, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15221670144, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15224029440, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15231959296, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15232647424, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15235006720, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15235694848, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15243624704, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15251554560, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15259484416, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15260172544, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15268102400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15270461696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15278391552, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15280750848, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15281438976, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15289368832, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15290056960, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15292416256, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15300346112, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15308275968, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15316205824, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15318565120, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15326494976, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15334424832, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15335112960, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15343042816, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15343730944, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15344419072, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15345107200, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15353037056, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15360966912, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15363326208, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15371256064, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15373615360, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15375974656, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15376662784, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15384592640, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15386951936, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15389311232, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15397241088, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15405170944, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15405859072, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15413788928, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15421718784, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15422406912, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15430336768, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15438266624, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15446196480, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15446884608, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15454814464, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15457173760, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15459533056, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15460221184, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15462580480, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15470510336, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15471198464, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15479128320, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15479816448, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15487746304, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15490105600, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15490793728, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15498723584, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15506653440, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15514583296, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15515271424, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 15523201280, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15531131136, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15533490432, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15534178560, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15542108416, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15544467712, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15552397568, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15560327424, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15562686720, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15570616576, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15572975872, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15573664000, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15581593856, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15589523712, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15597453568, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15605383424, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15606071552, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15606759680, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15614689536, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15622619392, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15624978688, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15632908544, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15640838400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15648768256, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15656698112, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15657386240, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15665316096, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15667675392, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15675605248, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15683535104, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15691464960, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15692153088, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15700082944, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15708012800, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15715942656, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15723872512, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15731802368, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15734161664, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15736520960, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15737209088, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15745138944, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15753068800, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15760998656, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15761686784, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15764046080, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15771975936, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15779905792, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15780593920, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15781282048, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15789211904, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15791571200, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15793930496, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15794618624, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15795306752, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15803236608, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15803924736, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15811854592, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15814213888, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15822143744, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15824503040, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15832432896, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15840362752, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15848292608, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15856222464, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15858581760, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15859269888, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15861629184, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15862317312, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15863005440, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15870935296, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15871623424, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15879553280, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15880241408, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15888171264, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15888859392, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15891218688, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15893577984, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15901507840, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15909437696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15917367552, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15925297408, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15925985536, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15933915392, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15941845248, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15942533376, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15950463232, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15951151360, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15959081216, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15961440512, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15969370368, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15971729664, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15974088960, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15976448256, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15977136384, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15977824512, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15978512640, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15986442496, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 15994372352, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16002302208, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16004661504, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16007020800, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16007708928, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16010068224, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16017998080, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16025927936, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16026616064, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16034545920, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16042475776, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16050405632, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 16058335488, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16059023616, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16059711744, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16062071040, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16062759168, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16070689024, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16078618880, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16080978176, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16088908032, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16089596160, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16091955456, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16099885312, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16107815168, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16110174464, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16118104320, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16118792448, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16126722304, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16134652160, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16142582016, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16143270144, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16151200000, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16159129856, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16159817984, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16167747840, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16175677696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16183607552, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16185966848, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16188326144, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16189014272, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16189702400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16192061696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16199991552, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16202350848, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16203038976, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16210968832, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16218898688, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16219586816, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16227516672, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16229875968, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16237805824, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16240165120, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16242524416, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16250454272, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16251142400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16251830528, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16252518656, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16254877952, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16257237248, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16257925376, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16260284672, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16262643968, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16270573824, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16278503680, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16280862976, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16281551104, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16283910400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16286269696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16288628992, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16290988288, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16291676416, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16294035712, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16294723840, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16302653696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16305012992, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16307372288, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16315302144, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16323232000, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16325591296, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16326279424, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16326967552, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16327655680, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16330014976, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16330703104, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16331391232, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16339321088, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16347250944, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16349610240, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16351969536, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16354328832, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16355016960, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16355705088, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16356393216, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16357081344, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16365011200, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16365699328, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16368058624, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16370417920, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16378347776, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16380707072, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16383066368, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16383754496, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16386113792, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16388473088, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16390832384, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16391520512, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16393879808, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16396239104, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16398598400, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16406528256, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16414458112, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16415146240, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16423076096, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16423764224, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16424452352, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16426811648, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16429170944, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16429859072, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16437788928, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16445718784, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16446406912, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16454336768, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16455024896, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 16462954752, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 16470884608, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 16471572736, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 16473932032, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 16476291328, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 16484221184, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 16492151040, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16492839168, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16500769024, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16501457152, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16509387008, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16510075136, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16510763264, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16518693120, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16519381248, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16527311104, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16535240960, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16535929088, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16536617216, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16544547072, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16552476928, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16560406784, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16568336640, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16569024768, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16569712896, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16577642752, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16580002048, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16587931904, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16588620032, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16596549888, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16598909184, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16606839040, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16607527168, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16608215296, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16616145152, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16624075008, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16624763136, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16625451264, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16627810560, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16635740416, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16638099712, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16640459008, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16648388864, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16656318720, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16664248576, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16664936704, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16667296000, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16675225856, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16677585152, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16685515008, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16686203136, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16694132992, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16702062848, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16704422144, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16712352000, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16720281856, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16722641152, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16723329280, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16725688576, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16733618432, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16734306560, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16742236416, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16750166272, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16752525568, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16760455424, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16762814720, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16765174016, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16773103872, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16781033728, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16783393024, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16785752320, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16793682176, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16796041472, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16796729600, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16799088896, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16807018752, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16807706880, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16810066176, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16817996032, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16818684160, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16819372288, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16821731584, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16824090880, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16826450176, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16827138304, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16835068160, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16842998016, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16843686144, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16844374272, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16846733568, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16849092864, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16857022720, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16864952576, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16865640704, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16868000000, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16875929856, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16883859712, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16891789568, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16892477696, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16900407552, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16902766848, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16905126144, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16905814272, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16908173568, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16908861696, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16909549824, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16917479680, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16925409536, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16926097664, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16928456960, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16929145088, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16929833216, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16932192512, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16940122368, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16948052224, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16948740352, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16949428480, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16951787776, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16954147072, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16962076928, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16962765056, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16970694912, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16971383040, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16972071168, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16974430464, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16976789760, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16977477888, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16979837184, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16982196480, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16990126336, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16992485632, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 16994844928, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17002774784, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17010704640, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17013063936, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17013752064, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17021681920, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17029611776, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17030299904, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17032659200, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17040589056, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17048518912, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17050878208, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17058808064, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17059496192, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17067426048, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17069785344, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17077715200, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17080074496, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17082433792, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17083121920, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17083810048, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17084498176, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17086857472, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17089216768, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17091576064, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17093935360, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17101865216, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17109795072, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17110483200, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17112842496, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17113530624, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17121460480, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17123819776, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17124507904, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17132437760, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17140367616, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17141055744, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17143415040, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17145774336, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17148133632, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17148821760, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17156751616, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17159110912, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17159799040, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17160487168, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17168417024, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17170776320, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17178706176, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17186636032, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17188995328, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17189683456, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17197613312, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17199972608, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17207902464, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17215832320, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17223762176, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17231692032, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17232380160, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17240310016, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17248239872, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17250599168, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17258529024, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17259217152, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17267147008, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17275076864, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17283006720, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17290936576, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17293295872, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17301225728, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17309155584, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17309843712, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17317773568, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17325703424, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17333633280, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17341563136, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17343922432, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17346281728, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17354211584, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17356570880, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17357259008, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17359618304, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17367548160, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17368236288, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17368924416, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17369612544, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17377542400, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17385472256, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17386160384, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17386848512, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17394778368, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 17402708224, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } } ]